Skip to content

LangNarrationOverride

LangNarrationOverride<L> = ElevenLabsVoiceConfig & object | { accent?: string; modelType?: "expressive"; name: BuiltInVoiceName<L>; pacing?: string; seed?: number; style: string; } | { accent?: never; modelType?: Exclude<ModelType, "expressive">; name: BuiltInVoiceName<L>; pacing?: number; seed?: number; style?: never; }

Defined in: voiceConfig.ts:82

Per-language voice override. Can override the default voice name and set a seed for TTS generation.

The voice name discriminates provider-specific settings: built-in model voices use expressive/consistent controls, ElevenLabs voices use the numeric eleven_multilingual_v2 controls.

L extends Lang = Lang

ElevenLabsVoiceConfig & object


{ accent?: string; modelType?: "expressive"; name: BuiltInVoiceName<L>; pacing?: string; seed?: number; style: string; }

optional accent?: string

Accent description for expressive synthesis. The more specific, the better — e.g. 'Southern American English' or 'Received Pronunciation British'. Omitted from the prompt when not set — the voice uses its natural default.

optional modelType?: "expressive"

Can be omitted when style is set — expressive is implied. Business tier only.

name: BuiltInVoiceName<L>

optional pacing?: string

Pacing description for expressive synthesis. Describes the overall speed and tempo — e.g. 'Measured and deliberate' or 'Brisk and energetic'.

optional seed?: number

Integer seed included in the audio cache key. A different seed always forces regeneration. Consistent output is not guaranteed across all voice types.

style: string

Speaking style prompt for expressive synthesis. Business tier only. Implies expressive model type.


{ accent?: never; modelType?: Exclude<ModelType, "expressive">; name: BuiltInVoiceName<L>; pacing?: number; seed?: number; style?: never; }

optional accent?: never

optional modelType?: Exclude<ModelType, "expressive">

TTS model type — modelTypes.expressive or modelTypes.consistent. Defaults to consistent.

name: BuiltInVoiceName<L>

optional pacing?: number

Speaking rate for consistent synthesis. Valid range: 0.25 to 2.

optional seed?: number

Integer seed included in the audio cache key. A different seed always forces regeneration. Consistent output is not guaranteed across all voice types.

optional style?: never